Can it be done?: Historical US production

[Secret Master]

I disagree.

I've done it with the USA.

Hell, I've built Stalin's insane naval plan from the 1930s and still beat the Axis in the war.
http://[URL="https://forum.paradoxp... will be) the point to play as Allies ?[/URL]
I've also built up the USA to the point where she has more effective IC than the rest of the world combined.

I have also built the historical Soviet OOB in time for Barbarossa in 1941. It's difficult, but doable under normal difficulty.

Getting the Red Army up to speed properly during the actual war is a more difficult proposition, because so much manpower is locked in special events/decisions. I wouldn't rule it out completely, though. If you don't lose too much IC or manpower to occupation, a competent Soviet player should be able to get damn close to historical Soviet mobilization. The problem in HOI3 is the same as HOI4, though. How do you track losses when trying to determine production?

 

[FOARP]

I like this exercise. From my point of view HOI4 is inferior in every aspect of other games of the series but in production.

You're seriously saying that HOI4 is inferior to HOI1?

 

[Indigo Blue]

I've done it with the USA.

Hell, I've built Stalin's insane naval plan from the 1930s and still beat the Axis in the war.
I've also built up the USA to the point where she has more effective IC than the rest of the world combined.

I have also built the historical Soviet OOB in time for Barbarossa in 1941. It's difficult, but doable under normal difficulty.

Getting the Red Army up to speed properly during the actual war is a more difficult proposition, because so much manpower is locked in special events/decisions. I wouldn't rule it out completely, though. If you don't lose too much IC or manpower to occupation, a competent Soviet player should be able to get damn close to historical Soviet mobilization. The problem in HOI3 is the same as HOI4, though. How do you track losses when trying to determine production?

Ah I did think of a different way to do your calculations if you abstained from war. Instead of all the approximations just build 1 thing while building your mic i.e. Artillery I. After you've built all the way to say 1944 while building MIC you can take your total art produced and divide out the number of unmodified MIC production days you used. So for example, lets say I end up with
600,000 Art I

That turns into 1,500,000 MIC production units which would be 500,000 MIC production days at 100% production efficiency and no throughput (base 5) from there you can allocation MIC production days to generate whatever battalions you were interested in that might compose accurate battle estimates.

I.e. a 7 inf 2 art no support would be 700/5 + 2*3.5*48/5 or 120 + 67.2 production days or ~200

 

[Dalwin]

Ah I did think of a different way to do your calculations if you abstained from war. Instead of all the approximations just build 1 thing while building your mic i.e. Artillery I. After you've built all the way to say 1944 while building MIC you can take your total art produced and divide out the number of unmodified MIC production days you used. So for example, lets say I end up with
600,000 Art I

That turns into 1,500,000 MIC production units which would be 500,000 MIC production days at 100% production efficiency and no throughput (base 5) from there you can allocation MIC production days to generate whatever battalions you were interested in that might compose accurate battle estimates.

I.e. a 7 inf 2 art no support would be 700/5 + 2*3.5*48/5 or 120 + 67.2 production days or ~200

You are losing me with this. Each items lists a production cost (I am not referring to the resources). Each factory produces 5 points toward that cost at 100% efficiency.

Those items combine to give you an exact cost needed to reach any specified OOB if everything started at 100%. The amount of time it takes to get from a new production line to 100% is also linear and is easily measured and taken into account. You can calculate a consistent rate for new MIC production (though I suggest if consumer goods is at 20% you then make every 5th new factory be CIC instead of MIC).

This should all combine into an algebraic equation which lists the total manufacturing capacity needed to reach a given OOB vs a second calculation for the total production that can be achieved at any given date.

Nothing is indefinite in this unless you want to factor in combat losses or conquered industry. In the latter case one would need to make some assumptions on how much was conquered when, but it still simply boils down to pure calculation with zero unknowns involved.

The biggest assumption to make (and an assumption is better than an anecdote based on a play through) is at what dates the US can change economic laws so that its new factory construction rate can increase.

 

[Indigo Blue]

You are losing me with this. Each items lists a production cost (I am not referring to the resources). Each factory produces 5 points toward that cost at 100% efficiency.

Those items combine to give you an exact cost needed to reach any specified OOB if everything started at 100%. The amount of time it takes to get from a new production line to 100% is also linear and is easily measured and taken into account. You can calculate a consistent rate for new MIC production (though I suggest if consumer goods is at 20% you then make every 5th new factory be CIC instead of MIC).

This should all combine into an algebraic equation which lists the total manufacturing capacity needed to reach a given OOB vs a second calculation for the total production that can be achieved at any given date.

Nothing is indefinite in this unless you want to factor in combat losses or conquered industry. In the latter case one would need to make some assumptions on how much was conquered when, but it still simply boils down to pure calculation with zero unknowns involved.

The biggest assumption to make (and an assumption is better than an anecdote based on a play through) is at what dates the US can change economic laws so that its new factory construction rate can increase.

You can do as described but again as you pointed out - and i'd also showed some of the math for it as well - even using a little calculus.

The suggestion i'm describing instead skips all the questions of ramp up or throughput efficiency. And instead focuses on the net IC that can be produced from the number of MIC that are generated. So instead of looking at each step of building we are looking at the final answer and then working backward to assess the number of IC days that have become accessible.

Think of it this way its the difference between a state function and non state function (if you know mathematics/physics/chemistry). State functions are basically descriptions of an endpoint without regard to how it was gotten there (my suggested analysis, just focusing on the total IC value of produced material). A non-state function would try to take into account how you got there (early analysis i suggested or your assessment) where we'd try to model the productivity of individual factories to get at the exact value.

To me, at least now, because secret master's problem for the most part is only looking at the start and end points than a state function analysis can fit. If he wants to take into consideration training it can be modeled by taking x divisions looking at their IC value and multiplying it by 6% and number of days trained to assess the amount of IC materials used.

 

[Dalwin]

You can do as described but again as you pointed out - and i'd also showed some of the math for it as well - even using a little calculus.

The suggestion i'm describing instead skips all the questions of ramp up or throughput efficiency. And instead focuses on the net IC that can be produced from the number of MIC that are generated. So instead of looking at each step of building we are looking at the final answer and then working backward to assess the number of IC days that have become accessible.

Think of it this way its the difference between a state function and non state function (if you know mathematics/physics/chemistry). State functions are basically descriptions of an endpoint without regard to how it was gotten there (my suggested analysis, just focusing on the total IC value of produced material). A non-state function would try to take into account how you got there (early analysis i suggested or your assessment) where we'd try to model the productivity of individual factories to get at the exact value.

To me, at least now, because secret master's problem for the most part is only looking at the start and end points than a state function analysis can fit. If he wants to take into consideration training it can be modeled by taking x divisions looking at their IC value and multiplying it by 6% and number of days trained to assess the amount of IC materials used.

On a completely separate side note: One problem with the example you gave for producing only artillery is that I have seen a bug report saying that if you stockpile more than 2 million of something the numbers go negative. This is a typical overrun error common in many computer programs. This is a bug that I would say realistically must be very low on the priority list to fix.

 

[Cardus]

You're seriously saying that HOI4 is inferior to HOI1?

HoI 1 was published about 14 years ago. If you take this in consideration and you compound it by the fact that 14 years of experience should have given a solid base for further developments I think that HoI 1 is better than HoI 4.

 

[Dalwin]

HoI 1 was published about 14 years ago. If you take this in consideration and you compound it by the fact that 14 years of experience should have given a solid base for further developments I think that HoI 1 is better than HoI 4.

I don't buy the "it's old so cut it slack" argument.

Taken to extremes one could then say that Pong was the greatest WWII simulation ever because without it the rest of the electronic game industry may not have followed.

 

[Cardus]

I don't buy the "it's old so cut it slack" argument.

Taken to extremes one could then say that Pong was the greatest WWII simulation ever because without it the rest of the electronic game industry may not have followed.

Well my argument was a little bit more sophisticated than that: when you make comparisons you need to compare apples with apples. Any game developed 14 years ago is inferior to a game developed today as well as any tank designed in the 1940s is inferior to a tank designed in the 1960s. But if you get first, in the 1940s, a design with sloped armor, Christie suspension, welded instead of riveted plates, etc. than hats off my friend.

To make a long story short: you need to adjust valuations depending on what is available at the time

 

[Secret Master]

Okay, so I did a test run. Not only did I fail, I failed hard. I didn't even get close. Let's look at some numbers. (Note: This test was on Veteran for reasons I will explain later.)

And here's the navy:

Let's talk about those factories for a second.

Since late 1943, the United States has had 100% of all factory slots filled at the highest IC techs (centralized industry, obviously); we are also sitting at Total Mobilization and extensive conscription. After I filled factory slots, I started converting CIC to MIC until I was down to 26 owned CIC; I can't convert anymore, or I will literally be unable to import the required rubber and aluminum (more on trade in a bit). I am receiving 78 CIC from exports even under Limited Exports (with no factory slots left, I am building RADAR, airfields, and other infrastructure all over the country; I don't even have room for nuclear reactors!). Note that CIC I receive in trade can't be used to import stuff, so my hands are tied in some respects.

The trade situation is a complete mess. I downshifted from Free Trade to Limited Exports sometime in 1942, because there is literally not enough aluminum in the world to fuel the MIC to even reach my meager numbers.
Take a look at this:

You will note that I have 220 MIC devoted to various aircraft, far more than anything else I'm producing. We are short 499 aluminum. I am also short rubber (there's 153 more to buy, but it doesn't make much difference at the moment for reasons that will be obvious). If you look closely, you can tell that my production lines are running as fast as possible (I didn't just switch them over for the screenshot) and they are building B-29s and Mustangs (Veteran difficulty). But I am short so much aluminum that even if I liberated Europe and was able to seize 100% of aluminum, I wouldn't be able to continue expanding production. Europe would just solve my current resource deficit. If I want to further increase production, there's nothing I can do to fuel it.

To put it another way, not only is it not possible for the USA to reach production goals, but no country on Earth could reach those production totals by 1945 even if they conquered the entire planet with max techs. You can put 600 factories on aircraft, and you won't be close to 96000 planes the United States produced just in 1944.

Trucks are another area that I couldn't even get close. I need 800,000 trucks by late 1945. There's plenty of oil in the world for the USA to tap, but rubber is finite. On Veteran, with max techs and max efficiency cap, I can produce 49.20 trucks per 15 factories per day. That means 15 factories would take 16260 days to reach 800,000. To get 800,000 trucks built in 4 and half years, I'd need around 150 factories producing trucks. That would require 2250 rubber and another 2250 oil. I don't think there's that much rubber and oil on the planet, even with max resource techs applied to all states.

I got much closer with tanks. I can produce 9.46 Shermans from 15 factories per day. I need another 30k tanks to reach my goal. 210 factories devoted to tank production would net me 26k tanks in 200 days. That's still not good enough. I'd need something closer to 240 factories running at full speed to get me to 60k by the time Japan is supposed to surrender.

I have 746k infantry equipment. I am currently producing 246 infantry kits per day per 15 factories (I have 30 currently tasked). Even if you accept that a 10:1 ratio for rifles to infantry kits (given that INF battalions only require 100 kits), I still need 150 factories producing infantry kits for 200 days to reach my goal of 12.5 million rifles (maybe more, depending on how certain things shake out with steel).

The navy is a different story. As you can see, I have plenty of carriers and battleships. I am short on DDs and subs. My current production puts me at 2.04 subs a week and 5.08 DDs a month. I obviously won't make the quota but with some better management and a few more NIC, it could be done. Note that I have 5400 total convoys, but around 500 are involved with trade. I am producing 2.18 a day. I have no idea if that is enough, but there is some slack in the system that can be taken out.

The question we have to ask ourselves is whether or not it makes sense to think of tanks and planes in terms of 10:1 or even 100:1 ratios to real world production. I know that HOI4 is not a simulator, but I'm sure I'm not the only one who would feel better if they could confidently say things like, "Sure, the USA is powerful enough. If you compute historical production using ratio X, you can see that she's where she needs to be, so stop whining." It might also put production from other countries into perspective.

EDIT: I ran this game on Veteran to see just how much impact PP had on the things.

 

[CavScout]

Interesting work.

 

[Dalwin]

Thanks for the effort and the results are interesting.

I do call foul on the trucks aspect however. There are a great many trucks abstracted into parts of the game. They are not all tied up in the points of motorized that are built. I am not sure that even half of the trucks get represented by motorized.

The supply system is undoubtedly simplified, but you do not and cannot assign motorized points to alter supply. This fact alone says that many of the war's trucks are abstracted.

 

[Axe99]

That's some impressive testing, great work and cheers for sharing . On the whole simulator thing, every model is a simulator, it's just what it's a simulation of. In the case of HoI, with semi-sensible numbers for armoured battalions (definitely) and air wings (I'm less confident about this, but it does look kinda like that, at least at the start), and quite good numbers for ships, it may well be a case that everything runs a bit better if equipment costs were less so that closer-to-historical numbers can be produced. One of the bigger issues with the AI is that it can't build enough equipment - perhaps the issue isn't the AI, but more the cost of the equipment?

If I didn't have a few months worth of naval ideas to work on, one thing I'd try is perhaps increasing the strength of the excavation techs substantially, so that the ramp up in resources generally follows the ramp-up in various factories (as it is, factories increase at I think more than twice the rate of resources, plus more factories are built, so a resource set that looks reasonable at the start of the game quickly runs out), along with halving the cost of all non-ship equipment, and see how that played out.

 

[Cardus]

Okay, so I did a test run. Not only did I fail, I failed hard. I didn't even get close. Let's look at some numbers. (Note: This test was on Veteran for reasons I will explain later.)
...

As I anticipated..

To get war production right you need a nonlinear production function that accounts for the increase of productivity. HOI3 somehow had that with praticals http://www.hoi3wiki.com/Practical_Knowledge

It seems from your test that the way in which production works in HOI4 you can't replicate the historical output.

 

[Meglok]

@Secret Master Very interesting, and thank you for doing this. I agree that something does seem to be off, but it may be we don't understand the coding conventions PDS used in creating the resource/manufacturing system. It would be very beneficial to hear from @podcat on this subject and your results.

 

[Dalwin]

Question: I presume one can mod the production cost of things (not talking about the resource cost). Cutting that in half for air would let everyone build twice as many planes. Yes they would still die in piles but that is rather beside the point.

 

[Meglok]

Question: I presume one can mod the production cost of things (not talking about the resource cost). Cutting that in half for air would let everyone build twice as many planes. Yes they would still die in piles but that is rather beside the point.

I think modifying the amount of rubber used in aircraft manufacture would be a good start. Burning 1 unit of rubber of each single and twin engine plane and 2 units for quads seems excessive.

 

[Dalwin]

I think modifying the amount of rubber used in aircraft manufacture would be a good start. Burning 1 unit of rubber of each single and twin engine plane and 2 units for quads seems excessive.

I don't disagree with you, but I think even if aluminum and rubber had not been short he could not have come close to building enough planes because of the non-resource production cost of aircraft. A full line of factories does not churn out enough planes per day.

 

[Axe99]

Question: I presume one can mod the production cost of things (not talking about the resource cost). Cutting that in half for air would let everyone build twice as many planes. Yes they would still die in piles but that is rather beside the point.

I think modifying the amount of rubber used in aircraft manufacture would be a good start. Burning 1 unit of rubber of each single and twin engine plane and 2 units for quads seems excessive.

Modding equipment costs is very straightforward, definitely doable. Cost for a piece of equipment is given by (this example is from a basic fighter, or a fighter level 2 in-game):

        build_cost_ic = 24
        resources = {
            oil = 1
            aluminium = 3
            rubber = 1
        }

I just gave it a quick test with destroyers, and the production system rounds to the nearest whole integer in terms of resource use after adding up all the fractions, so if you changed that rubber to 0.5, above, and used 10 factories to produce aircraft, it would only use 5 rubber (just don't use 10 lines of one factory each ), so that's definitely doable (and a good idea imo).


Edit: Maybe have a go at something like:

        build_cost_ic = 12
        resources = {
            oil = 1
            aluminium = 1
            rubber = 0.25
        }

Keep in mind that each equipment tier probably has its own production cost, so there's a little bit of cost editing involved. Recommend keeping a note in a comment next to the edited figure of the original.

 

[Secret Master]

If I didn't have a few months worth of naval ideas to work on, one thing I'd try is perhaps increasing the strength of the excavation techs substantially, so that the ramp up in resources generally follows the ramp-up in various factories (as it is, factories increase at I think more than twice the rate of resources, plus more factories are built, so a resource set that looks reasonable at the start of the game quickly runs out), along with halving the cost of all non-ship equipment, and see how that played out.

This could at least alleviate the resource crunch.

I suspect that the resource deficits are intentional, though. If I don't misread the design principles in play, it looks like the end goal is to make everyone, Axis and Allies and COMINTERN, feel resource shortages as time goes on.

The fact that we can continue producing stuff even when we are out of all resources required tells me that resources are meant to limit spamming of any kind of unit. This test clearly demonstrated that even if I could put 400 factories on a single piece of equipment, there's not enough resources in the world to support construction. You'll run out of oil, tungsten, steel, or rubber before all 400 MIC get in on the action.

As I anticipated..

To get war production right you need a nonlinear production function that accounts for the increase of productivity. HOI3 somehow had that with praticals http://www.hoi3wiki.com/Practical_Knowledge

It seems from your test that the way in which production works in HOI4 you can't replicate the historical output.

In terms of raw numbers, it clearly does not. But given levels of abstraction present in other ways (no trucks for moving supplies, convoys are multiple Liberty ships, 100 infantry kits per infantry battalion), it seems reasonable to rethink historical output in terms of ratios to in-game equipment. You can't produce 96,000 planes the entire game, let alone a single year, but perhaps 900 planes in HOI4 equals 90,000 planes historically.

It's something to consider, especially for modders.

As for nonlinear production increases, I will only say that there were some real advantages to HOI3's system, but boy did it encourage some weird behavior at times.

"Sure, let's build a 1918 battlecruiser."

"But it's a waste, comrade."

"I don't care. Stalin wants an ocean going navy, and we need practical knowledge to get him that navy. Now go build that crappy battlecruiser."

"Yes, comrade."

@Secret Master Very interesting, and thank you for doing this. I agree that something does seem to be off, but it may be we don't understand the coding conventions PDS used in creating the resource/manufacturing system. It would be very beneficial to hear from @podcat on this subject and your results.

If I thought it was possible without causing errors from losses, I'd run a parallel game with a human USA and human Germany with historical conquests (but not real fighting) and see how Germany's numbers shake out against US numbers.

I'm sure Germany would post better numbers in ratio to US numbers than historically, but it would be interesting to see how they look.